Download Time Series Forecasting Of Nifty Stock Market Using Weka

Time Series Forecasting Of Nifty Stock Market Using
Weka
Raj Kumar1, Anil Balara2
1
M.Tech, Global institute of Engineering and Technology ,Gurgaon
2
Associate Professor, Global institute of Engineering and Technology,Gurgaon
Abstract:
Forecasting accuracy is the most important factor in
selecting any forecasting methods. Research efforts
in improving the accuracy of forecasting models are
increasing since the last decade. The appropriate
stock selections those are suitable for investment is a
very difficult task. The key factor for each investor is
to earn maximum profits on their investments.
Numerous techniques used to predict stocks in which
fundamental and technical analysis are one among
them. In this paper latest time series forecasting
feature of Weka is used. Results has been checked
using for different methods. These methods applied
on 3 months NIFTY data. 10 days result has been
predicted. The results from SMO regression shows
that it offers the ability to predict the stock prices
more accurately than the other existing techniques.
The results will be used to analyze the stock prices
and their prediction in depth in future research
efforts.
1.Introduction:Generally, stock market across the globe replicates
the fluctuation of the market’s economy, and attracts
the attention of millions of investors. The stock
market is characterized by high risk and high yield;
hence investors are concerned about the analysis of
the stock market and are trying to forecast the trend
of the stock market. To accurately predict stock
market, various prediction algorithms and models
have been proposed in the literature. Forecasting is a
process that produces a set of output with a set of
variables [9]. The variables are normally past data of
the market prices. Basically, forecasting assumes that
future values are based, at least in part on these past
data. Past relationships can be derived through the
study and observation. Neural networks have been
used to perform such task .The ideas of forecasting
©
using neural network is to find an approximation of
mapping between the input and output data through
training. Later to predict the future stock value
trained neural network can be used [8].
The rest of the paper is organized as follows. Section
2 reviews the literature in predicting the stock market
price through various methods. Section 3 focuses on
the objectives of the research. Section 4 discusses
about the National Stock Exchange (NSE), stock
market, stock classification, forecasting, fundamental
analysis and technical analysis. Section 5 explains
the methodology of using Weka tool to forecast the
stock prices using time series forecasting package.
Section 6 draws the conclusion of the paper.
2.Literature reviw :Prediction of stock price variation is a difficult task
and the price movement behaves more like a random
walk and varies with time. Since the last decade,
stockbrokers and future traders have relied upon
various types of intelligent systems to make trading
decisions
Enke, D., Thawornwong [1] has been explained the
techniques of data mining and neural network needed
for prediction of stock market prices and values. It
has been widely accepted by many studies that nonlinearity exists in the financial markets and that
neural networks can be effectively used to uncover
this relationship .Wang, J.L., Chan [2] uses the daily
stock prices of Microsoft, Intel, and IBM to assess
stock market purchasing opportunities with simple
technical indicators. Sapna Jain[3] has been
assimilated the knowledge about Weka tool and
demonstrated with taking example of clustering
method and introduce WEKA workbench, reviews
the history of the project and k-means clustering
execution in WEKA 3.7. Hazem M. El-Bakry[4] a
new technique for fast forecasting of stock market
prices is presented. Such algorithm uses new high
speed time delay neural networks (HSTDNNs).
Swasti Singhal [5] has explained the basic concept
is used to analyze the data from different angle,
JRPS International Journal for Research Publication & Seminar Vol 05 Issue 02 March -July 2014
1
categorize it and finally to summarize it. They
introduce the key principle of data pre-processing,
classification, clustering and introduction of WEKA
tool. Weka is a data mining tool. They has been
described the steps of how to use WEKA tool for
these technologies. It provides the facility to classify
the data through various algorithms.
markets. Jageshwer Shriwas[6] explains the hybrid
approach in stock price prediction. this research
determines the feasibility and practicality of using
data
mining
as
a
forecasting
tool.
K.K.Sureshkumar[7] use Weka 3.6.3 tool to obtain
more accurate stock prediction price and to compare
them with weka classifier functions such as Gaussian
processes, isotonic regression, least mean square,
linear regression, multilayer perceptron, pace
regression, simple linear regression and SMO
regression.Most of these uses data mining prediction
with weka older versions. Very little is done on
forecasting package of weka on the Indian stock
markets
3.Objective of the study:The main objective of this paper is to use Weka 3.7.8
tool to obtain more accurate stock prediction price
using time series forecasting package of weka. This
research will examine and analyze the use of weka as
a forecasting tool. Specifically some mining and
forecasting ability to predict future trends of Stock
Market Indices will be tested.Our objective is to
analyze prediction of various methods on 3 months
record of NIFTY.Analyze and compare results by
plot graphs. We will forecast 10 days result using all
four methods.
4. National Stock Exchange of india
(NSE) :NSE was incorporated in November 1992, and
received recognition as a stock exchange under the
Securities Contracts (Regulation) Act, 1956 in April
1993. Since its inception in 1992, NSE of India has
been at the vanguard of change in the Indian
securities market. This period has seen remarkable
changes in markets, from how capital is raised and
traded, to how transactions are cleared and settled.
The market has grown in scope and scale in a way
that could not have been imagined at that time.
Indian equity markets are today among the most deep
and vibrant markets in the world. NSE offers a wide
range of products for multiple markets, including
equity shares, Exchange Traded Funds (ETF),
©
Mutual Funds, Debt instruments, Index futures and
options, Stock futures and options, Currency futures
and Interest rate futures. The role of trading members
at NSE is to the extent of providing only trading
services to the investors; the Exchange involves
trading members in the process of consultation and
participation in vital inputs towards decision making
[10].
4.1 Stock Market
A stock market index is a method of measuring a
stock market as a whole. The most important type of
market index is the broad-market index, consisting of
the large, liquid stocks of the country. In most
countries, a single major index dominates
benchmarking, index funds, index derivatives and
research applications. In addition, more specialized
indices often find interesting applications. In India,
we have seen situations where a dedicated industry
fund uses an industry index as a benchmark. In India,
where clear categories of ownership groups .exist, it
becomes interesting to examine the performance of
classes of companies sorted by ownership group.
4.2 Stock Classification
Stocks are often classified based on the type of
company it is, the company’s value, or in some cases
the level of return that is expected from the company.
Below is a list of classifications which are generally
known to us Growth Stocks, Value Stocks, Large
Cap Stocks, Mid Cap Stocks, and Small Cap Stocks.
Stocks are usually classified according to their
characteristics. Some are classified according to their
growth potential in the long run and the others as per
their current valuations. Similarly, stocks can also be
classified according to their market capitalization.
S&P CNX NIFTY has NIFTY (50), Junior NIFTY
(50), CNX IT (20), Bank NIFTY (12), NIFTY
Midcap50, CNX Realty (10) and CNX Infra (25).
The sectoral distribution of NSE are Financial
services or banks, Energy, Information Technology,
Metals, Automobile, FMCG, Construction, Media &
Entertainment, Pharma, Industrial Manufacturing,
Cement, Fertilizers & Pesticides, Textiles, Power and
Telecom [10].
4.3 Limitation in earlier version of Weka
Upto weka 3.7.5, forcasting feature was not
supported. Only clustering,classify and associate was
possible. Now version 3.7.8 and higher supports
JRPS International Journal for Research Publication & Seminar Vol 05 Issue 02 March -July 2014
2
forecasting feature which can predict values on the
basis on history[11].
levels.This technique is more flexible and less time
consuming.
4.4 Time Series Forecasting
Time series analysis is the process of using statistical
techniques to model and explain a time-dependent
series of data points. Time series forecasting is the
process of using a model to generate predictions
(forecasts) for future events based on known past
events. Time series data has a natural temporal
ordering - this differs from typical data
mining/machine learning applications where each
data point is an independent example of the concept
to be learned, and the ordering of data points within a
data set does not matter. Examples of time series
applications include: capacity planning, inventory
replenishment, sales forecasting and future staffing
Figure 1 :- Forecasting screen
5. Research Methodology:5.1Data Preprocessing
We will apply following four methods on 3
months data of nifty stock exchange.These data
minig methods are :
1.
2.
3.
4.
Gaussian processes
linear regression
Multilayer Perceptron
SMOreg
Statistical Value of Nifty
Std
De
v
Dist
inct
Uniq
ue%
677
9.79
308
60
100
756
3.5
682
4.07
314
.39
60
100
621
2.25
730
2.6
674
1.2
289
.81
60
100
622
1.45
736
7.1
678
3.31
295
.29
59
97
Min
.
Ma
x.
Me
an
Open
621
6.75
742
8.75
High
627
7.75
Low
Close
Attri
butes
©
In data preprocessing, we have consider 90 days
of CNX NIFTY daily data. From this, minimum
price, maximum price, mean, standard deviation,
distinct and unique percentage has been found.
The details are given in the table 1. From the
table 1, we found the Minimum, Maximum,
Mean and Standard deviation for the four
attributes. In view of the data, we find out inbetween values for the stock prices. In table 2
we have classifications in the table and in the
graphs each in four categories namely open
price, high price and low price,close price. For
example, consider the Table 2 open column
containing the attribute data as 13 which means
13 days of data indicates stock price range lies
between 6216.75 and 6519.75. Similarly for data
29,7,11. The last data 11 which denotes 11 days
of Nifty
stock price range lies between
7125.75and 7428.75. Totally data of 60
instances of open price, high price and low
price and close price are represented in the table
2 and 3 and lies within value of minimum and
maximum.
JRPS International Journal for Research Publication & Seminar Vol 05 Issue 02 March -July 2014
3
OPEN (6216.75-7424.75)
Data
Lies Between The Price
HIGH(6277.75-7563.5)
Data
Lies between The Price
13
6216.75-6519.75
8
6277.75-6534.9
29
6519.75-6822.75
29
6534.9-6792.05
7
6822.75-7125.75
9
6792.05-7049.2
11
7125.75-7428.75
6
7049.2-7306.35
8
7306.35-7563.5
Table 1. Statistical values of NIFTY
Table 2. Data Classification of Open price, High Price
LOW(6212.25-7302.6)
Data
Lies Between The Price
CLOSE(6221.45-7367.1)
Data
Lies between The Price
9
6212.25-6484.83
9
6221.45-6507.86
31
6484.83-6757.42
30
6507.86-6794.27
6
6757.42-7030.01
7
6794.27-7080.68
14
7030.01-7302.6
14
7080.68-7367.1
Table 3. Data Classification of Low Price,Close price
©
JRPS International Journal for Research Publication & Seminar Vol 05 Issue 02 March -July 2014
4
Figure 2 :- Data Preprocessing Graph of four classes (open.high.low,close)
5.2Results and Discussion
The all four methods Gaussian process, Linear
regression, Multilayer perceptron ,SMO
regression has been applied on 3 months of nifty
stock market.All methods are applied on 60
instances and the next 10 days of data has been
predicted.The forecast value and graph of 10
days data for smo method is shown in figure3
and figure 4.
The sample data has been tested on Windows 7
operating system. We have used Weka 3.7.8 tool
for preprocessing, evaluation and forecasting of
the stock . From this analysis, it is found that the
forecasting values of next 10 days results has
shown in the figures. We acquire four functions
for this analysis to forecast values and evaluate
them. By comparing the results of the direction
accuracy values and error percentage the SMO
regression is the best suited method for
forecasting the stock prices.Following are the
results .
Figure3:- 10 days forecasted result of
SMO regresion
SMO regression
©
JRPS International Journal for Research Publication & Seminar Vol 05 Issue 02 March -July 2014
5
Figure9:- Forecasting Graph of SMO regression
[5] Swasti Singhal, Monika Jena(2013),A Study
on WEKA Tool for Data Preprocessing,
Classification and Clustering, IJITEE.
Conclusion:
In this research, we examined and applied
different forecasting techniques by using the Weka
tool. we compared various prediction functions,
and found that SMO regression function offer the
ability to predict the stock price of NSE more
accurately than the other functions such as
Gaussian processes, linear regression, multilayer
perceptron. This analysis can be used to reduce the
error percentage in predicting the future stock
prices. It increases the chances for the investors to
predict the prices more accurately by reduced error
percentage and hence increased profit in share
markets. More Accurate results can be found if we
will take data of more duration. It will help to
check deviations in Values and predict more
accurate results.
Future work can be done by apply more
techniques and verify results with actual values.
Error rate and graph can be designed to study it’s
validity. After this step, by observing error rate on
different method, best technique can be selected.
All this steps can be repeated with different set of
data.
References:[1] Enke, D., Thawornwong, S. (2005) “The use of
data mining and neural networks for forecasting
stock market returns”, Expert Systems with
Applications, 29, pp. 927-940.
[2] Wang, J.L., Chan, S.H. (2006) “Stock market
trading rule discovery using two-layer bias
decision tree”, Expert Systems with Applications,
30(4), pp. 605-611.
[6]. Jageshwer Shriwas (2014), “Stock Price
Prediction Using Hybrid Approach of Rule Based
Algorithm and Financial News”, IJCTA..
[7].
K.K.Sureshkumar (2011),” An Efficient
Approach to Forecast Indian Stock Market
Price and their Performance Analysis”,IJCTA.
[8] Abhyankar, A., Copeland, L. S., & Wong, W.
(1997). “Uncovering nonlinear structure in realtime stock-market indexes: The S&P 500, the
DAX, the Nikkei 225, and the FTSE-100”.
Journal of Business & Economic Statistics, 15, 1–
14.
[9]. Atiya, A. F, E1-Shoura, S. M, Shaheen, S. I
and El-Sherif, M. S. “A comparison between
neural network forecasting techniques case study:
river flow forecasting,” in IEEE Transaction
Neural Networks pp. 402-409, 10-2, 1999.
[10] National Stock Exchange of India,
http://www.nseindia.com/content/us/fact2011_sec
1.pdf
[11] www.weka.wikispaces.com.
[12] J. G. Agrawal, Dr. V. S. Chourasia (2013),
“State-of-the-Art in Stock Prediction Techniques”,
IJAREEIE
.
[3] Sapna Jain, M Afshar Aalam(2010),“K-Means
Clustering Using Weka Interface”, Proceedings of
the 4th National Conference; INDIACom.
[4] Hazem M. El-Bakry, and Wael A. Awad
(2010), “Fast Forecasting of Stock Market Prices
by using New High Speed Time Delay Neural
Networks” , International Journal of Computer and
Information Engineering 4:2 2010. Pp 138-144.
©
JRPS International Journal for Research Publication & Seminar Vol 05 Issue 02 March -July 2014
6